clip.qa is a mobile app that turns screen recordings into structured bug reports. Record a bug on your phone, and clip.qa's AI drafts a report with steps to reproduce and device context for you to review. Reports export as markdown for AI coding tools like Claude Code and Cursor, or to paste into any tracker. Jira, Linear, Slack, GitHub Issues and Notion integrations are coming soon. No SDK required in the app you test.

What is Vibe QA — and why does it matter?

Vibe QA is quality assurance for apps built with AI coding tools — a practice that emerged alongside vibe coding (using AI assistants like Claude Code, Cursor, and Copilot to build apps rapidly). Because vibe-coded apps ship faster than traditional QA can keep up, Vibe QA uses AI-native tools to close the loop: capture bugs visually, generate LLM-ready reports automatically, and feed them back to the same AI tools that built the app. clip.qa is built for Vibe QA.

Does clip.qa work with OpenClaw projects?

Yes. clip.qa produces a structured report your agent, including OpenClaw, can read through MCP or a pasted prompt.

How is clip.qa different from Jam?

Jam has a Chrome extension, a Mac app and an iOS app, plus AI summaries and an MCP server. No Android app is documented. clip.qa records on iOS, Android and in Chrome (beta, request access), with no SDK in the app you test. clip.qa's MCP also records each fix attempt (pending, succeeded, failed) with its commit or PR, and can diff reports.

Do I need to install an SDK or add code to my app?

No. clip.qa works without any SDK integration. It uses your phone's native screen recording, so it works on any app — including competitor apps you're testing, OpenClaw projects, and apps you don't have source access to (subject to OS limits such as FLAG_SECURE).

How much does clip.qa cost?

Free: 30 bug reports and 30 AI-assisted reports a month. Team ($12.99/mo or $129/yr, billed per creator seat, viewers free): unlimited bug reports, 200 AI-assisted reports and 50 share links a month. Enterprise pricing is custom; confirm current quotas and features with sales. Audit logs and priority support are listed; SAML/SSO is coming soon.

What platforms does clip.qa support?

The clip.qa web dashboard is available now; the iOS app, Android app and Chrome extension are in beta (request access at hello@clip.qa). Bug reports can be copied as markdown into any tracker (Jira, Linear, GitHub Issues; native integrations coming soon) or used directly as prompts for Claude, Cursor, and other AI coding tools.

Codex

clip.qa / Use cases / Codex

clip.qa for Codex: send bug reports your agent can fix

Codex is built to work inside your repo. clip.qa gives it what a repo cannot show: a recording of the bug on a real device, turned into steps, screenshots and context your agent can read.

The clip.qa iOS app, Android app and Chrome extension are in beta. Request access on the apps page.

Request beta access See all integrations

The agent

About Codex by OpenAI

Codex is OpenAI's coding agent. The open-source Codex CLI runs in your terminal and edits your repository. Its docs list support for MCP servers over remote HTTP and over a local process, which is how it can read a clip.qa report without any copy and paste.

How it works

From a bug on your phone to a fix attempt in Codex

1. Record the bug on your own device

Use the clip.qa iOS or Android app (beta, request access), or the Chrome extension (beta, request access), to screen-record the bug. No SDK goes into the app you test.

2. Review the AI draft

clip.qa turns the recording into a report: title, severity, kind, suspected root cause, steps to reproduce, and expected vs actual behavior when the recording shows a defect, plus key-moment screenshots and device context (OS, device, memory, thermal). It is a draft, so check it before you hand it to Codex.

3. Send the report to Codex

Connect the clip.qa MCP server to Codex and ask it for the latest report, or copy the report as Markdown and paste it into a Codex session.

4. Review the diff and log the attempt

Codex gets the steps and context it needs to attempt a fix. You review the diff. Over MCP, your agent can also log each fix attempt (pending, succeeded or failed) with its commit or PR.

Connect Codex

Two ways to give Codex a clip.qa report

clip.qa has written setup guides for Claude Code, Cursor and Windsurf. Codex connects through the same standard MCP details. OpenAI did not build this connection.

clip.qa connection details

Remote MCP server:  https://mcp.clip.qa/mcp
Local MCP server:   npm @clip-qa/mcp-server
CLI:                npm @clip-qa/cli

1Add clip.qa as an MCP server in Codex using the remote endpoint https://mcp.clip.qa/mcp, or run the local npm package @clip-qa/mcp-server with your clip.qa API key. Follow the MCP page in the Codex docs for the exact config form.

2Prefer the shell? Codex runs in a terminal, so the clip.qa CLI (npm @clip-qa/cli) is another route.

3No setup at all: copy the report as Markdown from clip.qa and paste it into the Codex session.

Honest fit

Choose Codex on its own if

The bug reproduces from code or a failing test, with no device in the loop.
You already have the steps and logs written down.
You want a terminal agent for refactors and features, not for bug intake.

Codex writes and edits code. clip.qa does not. clip.qa is for the moment before that: you hit a bug on your phone in an AI-built app, and typing the steps from memory loses half of them. For solo developers and small teams shipping mobile apps, the recording becomes the report.

clip.qa and Codex FAQ

Does clip.qa work with Codex?

Codex documents MCP support for remote HTTP and local servers, so it can connect to https://mcp.clip.qa/mcp or to the @clip-qa/mcp-server package. clip.qa has setup guides for Claude Code, Cursor and Windsurf; for Codex use the generic MCP details above. You can also paste the report as Markdown.

Will Codex fix the bug for me?

Codex gets the steps, screenshots and device context it needs to attempt a fix. It does not fix every bug, so review the diff the way you would review any pull request.

Is this an OpenAI integration?

No. clip.qa is an independent product. Codex is built by OpenAI, and there is no partnership. The connection uses the standard MCP protocol that Codex supports.

Codex edits the code. clip.qa captures the bug.

Record the bug, review the AI draft, then send it to your agent. Free plan available.

Request beta access See all integrations